This is the government. Loss of reality, no common sense, and grand delusions about their own skill and capabilities is standard.
It is government workers. You know, the stupidest of the stupidest combined with top scores in arrogance.
Also provides an useful indicator whether your overclocked rig is running too hot!
This is the only thing that will get us better enterprise IT security. Insurances actually care about not paying, and hence they care about actual risk. They do not care about "compliance" unless it actually decreases their risk.
(Side note: Anything you cannot get insured, like a nuclear reactor, is a very bad idea in the first place.)
The reason these small sample sizes are used is that with decent sample sizes you would not find all those effects that allow you to publish a lot. That is the whole problem.
I was taught that the scientific method welcomed challenges to accepted beliefs - a return to that position would go a long way towards reforming belief in science.
The method does, but most scientists do not, as they routinely oversell their results and challenges could actually endanger their funding or reputation. If we were to select professors and researchers actually on scientific merit, and not on "best show provided", things may be different. But today doing good science is a sure way to not be able to work as a researcher beyond a PhD, and even finishing that PhD can be tricky, as you will not publish enough. The system is completely borked because the wrong people have been promoted for a long, long time.
We are well advanced on that path....
Not only measurements. There is a lot of defective CS research being published in all CS fields. "Good" conferences have the additional disadvantage that the trash submitted there is harder to spot, as it is well written trash.